Re: Visualising comparative programming language performance

[Jerome Vuarand <jerome.vuarand@...>]

2009/6/1 Doug Rogers <doug.rogers@elbitsystems-us.com>:
> David Given wrote:
>>
>> Apropos of nothing much, people might be interested in the following link:
>> http://gmarceau.qc.ca/blog/2009/05/speed-size-and-dependability-of.html
>
> That was very interesting. Marceau technique seems to capture meaningful
> data in a visually attractive way. Thanks for the link!
>
> It looks like the biggest Lua/LuaJIT performance outlier in the 2009 test is
> fannkuch. The top performers take advantage of the multicore host system by
> creating threads to split up the work. It would be interesting to see if
> throwing luaproc at the problem would help. The test system is a quad core
> (mine at work and home are dual). A three-fold speed up would move
> Lua/LuaJIT up from rank 24 to 18 or so.
>
> I've used neither luaproc nor LuaJIT for anything more than an initial
> evaluation, so it will be a fun little project while I'm away over the
> weekend. If I can get it working, I'll report back. Others should feel free
> to beat me to it.

Some experiment to make use of multiple cores in benchmarks has been
conducted by people on the #lua IRC channel on freenode a couple
months ago. I don't remember how it ended up though, you can go there
ask them.